Train your old Brain?


brain on treadmill


As we grow old we all hope for some way to keep, and even sharpen our wits. The media understand this and excitedly publish claims, often based on very dodgy studies, that computerised “brain training” programs, or courses of mental exercises, sold for remarkable sums of money, will restore or even improve on our former youthful abilities. This week the Centre for Longevity at Stanford University and the Max Planck institute for Human Development in Berlin jointly published a statement that strongly questions these claims. It is signed by 78 eminent researchers in mental ageing

About time too. Over the last 30 years very many of us have worked very hard indeed, for no commercial return, to find ways to improve mental abilities in old age. My own results, and those of colleagues I admire and trust, have been so disappointing that I now feel that the failures of brain training programs are, theoretically, far more interesting than the outrageous claims that some venal salespersons make for them.

woman brain training


For over 150 years neuroscientists have known that some mental abilities are supported by particular, often quite tiny, specialised areas of the brain. A Japanese neuroscientist, Dr Ryuta Kawashima, was not the first to notice that when we carry out particular tasks the blood supply to the particular parts of the brain that support them briefly increases. (There is now good evidence that intense practice of particular skills does result in changes in the particular parts of the brain on which they depend. Still missing is evidence that practising particular skills results in changes in the entire brain that also benefit other skills that have not been trained).

Dr Kawashima was not the first to speculate that practising mental tasks might improve both local and general brain-blood circulation and so general brain health and mental competence. He was the first to develop a hand-held “Brain Training” games console that guides purchasers through simple tasks involving mental arithmetic, memory and decision speed and scores their performance in terms of estimated “Brain Ages”. I cannot find any way in which sensible estimates of “Brain Age” can be derived from first scores on any behavioural task. For the Kawashima tasks they are calculated from performance on first attempts when middle-aged and elderly purchasers, unused to hand-held game consoles, have initial difficulties and may be distressed by estimates of their current “brain ages. Happily they may soon become elated as their inevitable improvement is translated into to a gratifyingly young brain age.

Dr K brain age

One sad snag is that old and young people improve with practice at different rates. On   the simplest possible laboratory tasks it takes only a few hundred trials for most of us to become as fast as we can ever get. Even after their first scores of attempts young adults are almost as fast and accurate as they will ever become but elderly people improve markedly from a much lower initial level to a lower final peak. At any age, on any task, practice always improves performance but in old age we need more time and effort to reach lower peaks than we could once attain (1). It is encouraging for purchasers of brain-training devices to be told that their prolonged improvements are evidence that their brain ages are steadily becoming more youthful. I do not think that this is what is happening.

Boosts to elderly morale are always good and, if the tasks we learn are useful in our daily lives, even modest gains are worth the effort. It is an excellent idea to devise programmes to re-train older people, such as myself,  as fast and pleasantly as possible on any particular skills that make our lives more interesting and comfortable. It would be a greater achievement to invent training programs that do not just improve our competence on a single task but also on many others that have not been practised. Very many of us have tried very hard to do this and do not think that we have yet succeeded.

brain smart

An inconvenient reality is that on more difficult skills, such as languages, mathematics or chess, older people not only start from much lower baselines but improve much less, and far more slowly than the young do. Even after long practice they never reach the performance peaks that they once attained when young. The harder skills are the greater will be the differences in learning times and in peak achievements between old and young. Of course, even very modest improvements on life-enhancing skills, like languages or mental arithmetic, are immensely rewarding and useful but, alas, older people need more time and effort to achieve them.

Sales of Kawashima’s “Brain Trainer” exceeded 3 million units in a market in which sales of 1 million are exceptional. He was urged to make controlled trials to provide evidence that practice on his brain training tasks also generalised to improve performance on others. Ichiko Fuyomo (2), reviewing Kawashima’s work for the prestigious scientific journal Nature, reports that he declined to do this. Perhaps Kawashima was prudent. Other well-intentioned efforts to evaluate brain training programmes as tools to rehabilitate cognitive performance in older people have provoked surprisingly unpleasant disputes in our placid grey science.

What has become very ugly, but also very interesting is the vehemence with which those involved in marketing of these devices react to research that questions whether they are useful. Adrian Owen, with colleagues based in Cambridge and Manchester took advantage of an offer by the BBC to sponsor a web-based experiment for a TV program “Bang goes the Theory”. In 2010 they collected data from 11,430 mainly young and middle-aged people, who had trained themselves for about 3 sessions a week for 6 weeks on tasks that involved memory, planning, spatial skills and selective attention. They published their results as a paper titled “Putting Brain Training to the Test” (3) in which they reported that, of course, people of all ages became much better on all of the particular tasks that they practised. They found no evidence that these improvements extended to other, different, unpractised tasks, even when these were quite similar to those included in the original training battery. Immediately after their paper appeared the editor of “Nature” received a letter demanding its withdrawal from a distinguished North American scientist who had reported positive results with a different training program that is currently being marketed by a company for which he is scientific advisor.

In any science, disagreements between experiments are inevitable, and frequent. They are also essential for progress because differences between results always identify important differences in assumptions, in logic, in experiments or in analyses. Endless challenging of evidence is the core of the scientific method. Science advances by disagreements and is crippled by attempts to suppress results or opinions.

It is a fine ambition to discover training techniques that can quickly and painlessly make us better at particular skills. To discover “master tasks” that can also make us much better at everything else would be a far more wonderful thing and  has been a very old ambition for educationalists. Until the early 20th century, School and University curricula were based on the idea that this is just what a solid education in classical Greek and Latin achieves. A recent meta-analysis of results from 23 different studies of schoolchildren considered training programs designed to bring about general and lasting improvements in Working Memory. There was no evidence for such generalisation: any improvements were restricted to specific kinds of material (4). With many of my colleagues I have been frustrated to find that ability on any particular skill is surprisingly specific and often does not generalise even to other similar situations. Part of my personal experience of this came from supervising a study by Belinda Winder who based her Manchester MSc thesis on an experiment to test the conviction of many elderly Manchester volunteers that they had preserved their everyday competence by constantly practicing cryptic crossword puzzles. Belinda compared fanatical cryptic crossword experts aged between 70 and 85 against non-experts of their own age on memory tests, intelligence tests, vocabulary tests and tests of mental speed. She found that on all of these unfamiliar tests the experts were no better than the non-experts. Tim Salthouse reports similar results (5). It seems surprisingly difficult to show that lifelong expertise on one skill necessarily generalises to improve or preserve ability in other, quite similar, skills.

I now believe that this remarkable failure of general brain training is theoretically far more important than claims, based on dubious experiments, for general, overall mind improvements. I do not (yet) despair of becoming better at regaining some old competences, such as playing  chess better than I now do, or reading Russian and Latin or doing simple mental arithmetic. My personal research, and that of the colleagues I most respect and trust convinces me that I should not waste money on commercial training programs that promise to improve my performance even on other tasks that I do not practise. In my old age I am content to improve, as much as I can, on one thing at a time because I think that this is all I have ever been able to do. I also think that the reasons why this should be the case are very interesting indeed.


  1. Rabbitt, P.M.A. (1993). Crystal quest: A search for the maintenance of practised skills into old age. p 188-230 In Baddeley, A.D. & Weiskrantz, L., (eds) Attention, Selection Awareness and Control. A tribute to Donald Broadbent. Oxford, Clarendon Press, OUP.
  2. Fuyuno, (2007). Brain craze.Nature447 , 18-20.
  3. Owen, A. M., Hampshire, A., Grahn, J. A., Stenton, R., Dajani, S., Burns, A. S. et al. (2010). Putting brain training to the test.Nature,465, 775-778.
  4. Melby-Lervåg, M., & Hulme, C. (2013). Is working memory training effective? A meta-analytic review.Developmental Psychology, 49, 270.
  5. Hambrick, D. Z., Salthouse, T. A., & Meinz, E. J. (1999). Predictors of crossword puzzle proficiency and moderators of age–cognition relations. Journal of Experimental Psychology: General,128, 131-142.



About Gray Rabbitt

Grumpy gerontologist
This entry was posted in Uncategorized. Bookmark the permalink.

5 Responses to Train your old Brain?

  1. This is a very interesting article and I agree with the majority of it. In fact, I have just published a chapter on the myths of the use-it-or-lose-it Theory. The only aspect of this article which I do not agree with is the idea that increasing memory self-efficacy in old age is a good thing. My research has shown that an increase MemSE without an increase in actual cognitive functioning in older adults may result in a faster decline in cognition. This is because older adults become over-confident with their cognitive functioning and reduce the amount of cognitive effort they use for different tasks.

    I completely agree that there needs to be independent research into brain training. I have written a number of proposals for this type of research but have constantly been re-buffed. One must wonder if there is some kind of influence being exerted by companies who produce these programs on the funding councils. Owen et al. (2010), conducted research on adults who were all aged under sixty (if I remember correctly) and did not have an inactive control group. These two methodological issues are puzzling when Owen et al. (2010) was supposed to be testing brain training. I have also recommended using a within-subjects design (which I have used to test the effects of doing a cryptic crossword on cognitive functioning in older adults); once again, this idea has been snubbed for reasons which I cannot understand.

    A very interesting article and more work is needed!

    • Gray Rabbitt says:

      Thanks for this. Glad you found it useful. Pat Rabbitt

    • Gray Rabbitt says:

      Thanks for these useful and provocative comments. I cannot, ofcourse, think of any reason why your work has not found an outlet. I think that Owen et al did test some people over 60 but probably only a very few , if any, in their 70’s. On MemSE, your point is interesting but I seem to miss a point in your summary of your research: “My research has shown that an increase MemSE without an increase in actual cognitive functioning in older adults may result in a faster decline in cognition. This is because older adults become over-confident with their cognitive functioning and reduce the amount of cognitive effort they use for different tasks”. I am OK with the idea that false confidence may reduce scores on behavioural tests but puzzle that you should regard this as evidence of a decline in “cognition” or in “cognitive ability”. Over confidence may result in individuals trying to do things that they should realise that they cannot do – and so scoring a higher ratio of errors to correct responses but, unless it reduces the number of correct responses that they can make (which I do not think is the case) we cannot argue from this that their baseline cognitive ability has been impaired more than that of less over-weening individuals who have lower opinions of their own efficacy ?

      Anyhow, you are right , of course, that, as always!further work is required” best, Pat Rabbitt

      • Dr Nick Almond says:

        Thanks for your comment. My point is, if older people become over confident in their cognitive abilities then they will not put in as much cognitive effort into tests which could be quite demanding. This means that if a computer game is telling an older adults that their cognitive ability is getting better (when in reality it is just practise effects) then the older adult will not develop strategies for overcoming different cognitively challenging tasks. So this means that the older adults are not really using their metacognition system and when they actually realise that their cognitive function is not as good as it used to be then it is too late for them to re-engage their control pathway of metacognition.

        The point is when people were doing control tasks they had an increase in MemSE because their cognitive abilities were not being tested. However when the people did the experimental task (cryptic crossword) their MemSE decreased, but for the people who were poorer at cryptic crosswords their metacognition accuracy increased (but over the time period of the study their actual memory function did not improve). So this suggests that some older people are too confident in their cognitive functioning and therefore because they do not feel like they need to test it that much then they do not develop strategies that counter act the inevitable decline in cognitive abilities.

        I have just had another chapter accepted which describes the Use-It-Or-Lose-It theory in a model. When I get the PDF I will be happy to send you a copy if you wish?

        I hope I have explained what I meant about the relationship between MemSE and actual cognitive functioning / ability, if not let me know and I will try and explain it again. I have just submitted an article to healthy ageing research about this experiment. So if it is published I will let you know the reference… It is a free to view journal.

        Kind Regards


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s